1. Field of the Invention
The present invention relates generally to the processing of video images, and more particularly to techniques for deinterlacing video images.
2. Description of the Related Art
The world's major television standards use a raster scanning technique known as “interlacing.” Interlaced scanning draws horizontal scan lines from the top of the screen to the bottom of the screen in two separate passes, one for even numbered scan lines and the other for odd numbered scan lines. Each of these passes is known as a field.
It is often necessary and/or desirable to convert an interlaced video signal to a progressively scanned signal. A single image or frame of a progressively scanned video source is scanned from the top of the screen to the bottom in a single pass. Such a conversion is advantageous, for example, when using a progressive scan display. Many of the display devices used to show interlaced video signals are actually progressive in nature. Common examples of such progressive displays are computer CRTs and liquid crystal displays (LCDs). One advantage of such displays is that they eliminate many of the deficiencies of interlaced video formats, such as flicker, line twitter, and visible scan line structure. Another advantage of the progressive scan format is improvement in the compression ratio of a digital format, such as a digital broadcast or satellite transmission.
In order for such an interlaced video image to be utilized by a progressively scanned video system, it must be converted from an interlaced scan format to a progressive scan format. There are a number of standard techniques for performing this conversion, and range from simple merging of adjacent fields, to interpolation from a single field to form a progressive frame, to more complex motion-adaptive techniques.
Motion-adaptive deinterlacing of interlaced video sequences is a method which optimizes image quality by utilizing different deinterlacing techniques in image areas with and without motion. Where there is no motion in an image, pixel data from adjacent interlaced fields can simply be merged to form a progressive frame since the two fields are effectively temporally consistent. However, where motion exists in the image, two adjacent fields cannot simply be combined without generating interlace motion artifacts since the fields are not temporally consistent in those areas (with motion). Instead, new pixels must be calculated based on the temporally ‘current’ field so that the entire output image is temporally consistent. Motion-adaptive deinterlacing merges these two approaches by combining data from two fields at image locations where there is no motion, and calculating new pixel data at image locations where there is motion. To do this, motion in the image must be detected on a pixel by pixel basis in order to decide which deinterlacing technique to apply at each pixel location.
Prior art in this field makes use of a number of techniques to detect image motion. A generally used method is to compute the difference between two fields. A significant difference is classified as ‘motion’. One such method is to calculate the difference between spatially similar pixels in pairs of even fields and pairs of odd fields. (I.e., field pairs which are two field periods apart.) Another method calculates the difference between one field and a spatially coincident ‘virtual’ field which is created from a temporally adjacent but spatially non-coincident field. (I.e., the fields are one field period apart.)
The specific artifacts created by combining two fields with motion may also be detected. U.S. Pat. No. 5,625,421 discloses such a method, which detects “sawtooth” artifacts by testing sets of three vertically adjacent pixels for characteristics indicative of motion artifacts. Yet another method, disclosed in U.S. patent application Ser. No. 09/372,713, identifies motion artifacts by utilizing Fourier analysis to detect the presence of a specific vertical frequency which is characteristic of interlace motion artifacts. This prior art implementation of the identification of interlace motion artifacts looked only for a single frequency component. This frequency component is the maximum frequency (fmax) which can be represented by the vertical sampling rate of the video format—i.e., half the Nyquist sampling rate. The fmax frequency was identified by performing a partial Discrete Fourier Transform (DFT) which calculated only the specific frequency component of interest in local area around a given pixel location. Essentially, this partial DFT multiplied a vertical array of pixel values by the amplitude of a cosine wave of the fmax frequency, and then summed the results of the individual multiplications. The absolute value of this sum was used as the amplitude of the fmax frequency component.
While the latter frequency-based method does detect interlace motion artifacts, it also has a potential shortcoming in that it also erroneously detects high frequencies in certain static image features as ‘motion’. An example of this is illustrated in FIG. 1, which shows a simple example image 10 composed of a single rectangle 12 which has a different luminosity than the image background 14. The horizontally aligned borders 16 of the rectangle, when scanned in the vertical direction, appear as sharp transitions also known as a step function. An example of a step function is shown in FIG. 2, which also shows the frequency composition of such a step function. Although the step function is largely composed of lower frequencies 20, the specific frequency 22 which is characteristic of interlace motion artifacts is also present. Because this frequency is present in such images, it may be detected as a possible artifact, resulting in inaccurate detection. What is needed is a more accurate and reliable method of frequency detection which identifies interlace motion artifacts but rejects static image features which are not motion artifacts.